OpenAI fixes ChatGPT bug that made it overuse goblin and gremlin metaphors

After a rise in quirky creature metaphors in ChatGPT responses, OpenAI has fixed the issue by removing the training incentives that encouraged such language across model versions

Go to News Site